Skip to content
Xweather home

Xweather Met-in-a-box

A cutting-edge AI weather forecast ​for every country​

  • Outperforms traditional NWP
  • Trained on your local observations
  • Deploy on dedicated hardware, on-premises, or in the cloud

Build a custom AI weather model for your country, region, or operating area. It learns from your observations and runs on infrastructure you control — no supercomputer required.

Image of the Vaisala Xweather Met-in-a-box product. Sleek, dark computer box with ribbed front panel and small glowing status lights, isolated against a black background.

Vaisala Xweather is a selected partner for the most demanding organizations

  • DGAC Chile (Dirección General de Aeronáutica Civil)
  • Federal Aviation Administration (FAA)
  • Deutscher Wetterdienst (DWD)
  • Met Office
  • National Oceanic and Atmospheric Administration (NOAA)
  • Servicio Meteorológico Nacional (SMN) Argentina
  • U.S. Air Force
  • U.S. National Weather Service (NWS)
  • U.S. Navy
  • Logo and text reading Member of NBAA
  • DGAC Chile (Dirección General de Aeronáutica Civil)
  • Federal Aviation Administration (FAA)
  • Deutscher Wetterdienst (DWD)
  • Met Office
  • National Oceanic and Atmospheric Administration (NOAA)
  • Servicio Meteorológico Nacional (SMN) Argentina
  • U.S. Air Force
  • U.S. National Weather Service (NWS)
  • U.S. Navy
  • Logo and text reading Member of NBAA
  • DGAC Chile (Dirección General de Aeronáutica Civil)
  • Federal Aviation Administration (FAA)
  • Deutscher Wetterdienst (DWD)
  • Met Office
  • National Oceanic and Atmospheric Administration (NOAA)
  • Servicio Meteorológico Nacional (SMN) Argentina
  • U.S. Air Force
  • U.S. National Weather Service (NWS)
  • U.S. Navy
  • Logo and text reading Member of NBAA
  • DGAC Chile (Dirección General de Aeronáutica Civil)
  • Federal Aviation Administration (FAA)
  • Deutscher Wetterdienst (DWD)
  • Met Office
  • National Oceanic and Atmospheric Administration (NOAA)
  • Servicio Meteorológico Nacional (SMN) Argentina
  • U.S. Air Force
  • U.S. National Weather Service (NWS)
  • U.S. Navy
  • Logo and text reading Member of NBAA

National-scale forecasting without a supercomputer

Most countries and organizations rely on forecasts produced somewhere else, at a resolution and update cycle set by someone else.

Met-in-a-box gives you your own: a frontier AI model trained on observations from your territory, running in your environment, with data and forecasts under your control.

We can also design and install the observation network that feeds it, delivering a complete national forecasting system.

  • 100 ×

    higher resolution than traditional NWP models

  • Up to 50 %

    more accurate than traditional NWP models

  • 40,000 ×

    faster than traditional NWP models

How it works

01Observe

We connect all available observations, from weather stations and radar to upper-air soundings, and install new sensors where needed.

02Train

We train the AI model on decades of local data, teaching it how weather behaves over your mountains, coasts, and cities.

03Deploy

You decide where the trained model runs: on dedicated hardware, your existing infrastructure, or a cloud of your choice.

04Forecast

You get high-resolution forecasts in minutes, not hours, from a system under your full control.

Validation and proof

Proven partner to global weather institutions and defense forces

  • Decades of partnership: Meteorological institutions and defense forces have trusted Vaisala instrumentation since the 1930s.

  • National-scale AI deployment: The Philippines achieved modern AI forecasting capability by co-operating with Atmo.ai.

  • Institutional adoption: Vaisala hardware and data support leading agencies worldwide, including NOAA and the UK Met Office.

A man standing at a table in an anechoic chamber with white and black wall panels, surrounded by blue foam spires.
90 years

of measurement heritage

170 + countries

operating Vaisala instrumentation

300 + projects

delivered annually

Built for those who can't afford to guess

National meteorological services

Sharper forecasts, smaller budget

An AI model trained on your own weather, with no supercomputer to buy. We can also deliver the observation network that feeds it, up to a complete national forecasting system.

  • Up to 100× more detail and 40,000× faster than standard global forecasts

  • Trained on your observations

  • Observation network design and installation

Colorful heatmap of Earth focusing on Europe, Africa, and Asia, with ocean depths and geographic labels against a starry background.
Defense and civil protection

Weather intelligence under your control

When a storm is closing in or a mission is underway, your forecast can't depend on someone else's system.

  • Sovereign model and data

  • Compact enough to deploy where you operate

  • New forecasts in minutes, not hours

A drone flying against a dramatic sky at sunset, with clouds partially obscuring the orange and yellow hues.
Critical infrastructure

Forecasting for your exact area

Your assets sit in specific valleys, coastlines, and corridors. Met-in-a-box learns how weather behaves there, so forecasts reflect the local conditions around your sites and networks.

  • Sensors placed where your forecast needs them

  • Forecast detail that follows your terrain

  • Runs on infrastructure you control

An artistic map featuring a coastal region with stylized mountains, water bodies, and contour lines in varying shades of blue and green.
Abstract close-up of Vaisala Xweather Met-in-a-box, a metallic structure with vertical slats bending in a wave, lit by teal and copper light against a dark background.

National forecasting, now in a single device

Compact enough to take wherever you operate.

Frequently asked questions

A custom AI weather model for a country, region or operating area. It's trained on local observations and can be deployed on a compact system you control.

Traditional models solve the physics equations on a supercomputer, which takes hours. Met-in-a-box uses advanced frontier AI, trains on custom specifications, and provides the best possible solution available today. It delivers highly accurate forecasts in minutes — on far smaller hardware.

Leading forecast centers already rely on them. The European Centre for Medium-Range Weather Forecasts (ECMWF) has run its AI model in daily operations since February 2025, alongside its physics-based model. ECMWF reports gains of up to 20% on many measures, using about 1,000 times less energy. Met-in-a-box builds on physics-based weather science and over 90 years of Vaisala research and measurement technology. It combines that foundation with frontier AI and your observations.

You bring the observations you have. We'll help you place additional sensors where they add the most value.

You do. You decide who can access the system, where and how it’s deployed, and how its forecasts are used. Vaisala Xweather builds and supports the model, but your data and forecasts stay with you.

Yes. Alongside the model, we can design and install the observation network that feeds it. You decide how much we deliver and how much your team runs.

You choose the setup that fits your operations:

  • Dedicated hardware: a ready-to-run system accelerated by NVIDIA.

  • On-premises: on your existing computing infrastructure, inside your own facilities.

  • Cloud: in a cloud environment of your choice.

We're starting conversations with our first partners now. Talk to us to find out what Met-in-a-box could do for your territory.

Ready for your own AI weather model?